JargonFile/entries/one-banana problem.txt

one-banana problem

n. At mainframe shops, where the computers had operators for routine
administrivia, the programmers and hardware people tended to look down on the
operators and claim that a trained monkey could do their job. It was
frequently observed that the incentives that would be offered said monkeys
could be used as a scale to describe the difficulty of a task.

A one-banana problem is simple; hence, It's only a one-banana job at the
most; what's taking them so long? At IBM, folklore divided the world
into one-, two-, and three-banana problems. Other cultures had
different hierarchies and divided them more finely; at ICL, for
example, five grapes (a bunch) equals a banana. Their upper limit for
the in-house sysapes was said to be two bananas and three grapes
(another source claimed it was three bananas and one grape, but observed
however, this is subject to local variations, cosmic rays and ISO ).
At a complication level any higher than that, one asked the
manufacturers to send someone around to check things. See also
Infinite-Monkey Theorem.